CN113469158B - Method and system for identifying illegal hazardous chemical substance transport vehicle based on convolutional neural network - Google Patents
Method and system for identifying illegal hazardous chemical substance transport vehicle based on convolutional neural network Download PDFInfo
- Publication number
- CN113469158B CN113469158B CN202111035738.8A CN202111035738A CN113469158B CN 113469158 B CN113469158 B CN 113469158B CN 202111035738 A CN202111035738 A CN 202111035738A CN 113469158 B CN113469158 B CN 113469158B
- Authority
- CN
- China
- Prior art keywords
- chemical substance
- image
- vehicle
- classification
- dangerous chemical
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 239000000126 substance Substances 0.000 title claims abstract description 203
- 239000000383 hazardous chemical Substances 0.000 title claims abstract description 65
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 43
- 238000012549 training Methods 0.000 claims abstract description 54
- 238000012545 processing Methods 0.000 claims abstract description 46
- 238000004364 calculation method Methods 0.000 claims abstract description 44
- 238000013528 artificial neural network Methods 0.000 claims abstract description 11
- 238000002372 labelling Methods 0.000 claims description 28
- 230000005284 excitation Effects 0.000 claims description 17
- 238000004422 calculation algorithm Methods 0.000 claims description 13
- 238000004458 analytical method Methods 0.000 claims description 8
- 238000010606 normalization Methods 0.000 claims description 8
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 6
- 238000005520 cutting process Methods 0.000 claims description 5
- 238000003860 storage Methods 0.000 claims description 5
- 238000004590 computer program Methods 0.000 claims description 4
- 238000012797 qualification Methods 0.000 claims description 4
- 238000012795 verification Methods 0.000 claims description 3
- 230000032258 transport Effects 0.000 description 45
- 238000002790 cross-validation Methods 0.000 description 6
- 238000005094 computer simulation Methods 0.000 description 5
- 238000007726 management method Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 230000000007 visual effect Effects 0.000 description 5
- 239000002360 explosive Substances 0.000 description 4
- 238000013508 migration Methods 0.000 description 4
- 230000005012 migration Effects 0.000 description 4
- 238000011176 pooling Methods 0.000 description 4
- 239000003086 colorant Substances 0.000 description 3
- 238000012937 correction Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000013145 classification model Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000013523 data management Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000003709 image segmentation Methods 0.000 description 2
- 230000001629 suppression Effects 0.000 description 2
- 241000282414 Homo sapiens Species 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 210000004027 cell Anatomy 0.000 description 1
- 239000013626 chemical specie Substances 0.000 description 1
- 210000000078 claw Anatomy 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000003475 lamination Methods 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000010238 partial least squares regression Methods 0.000 description 1
- 231100000572 poisoning Toxicity 0.000 description 1
- 230000000607 poisoning effect Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 235000013547 stew Nutrition 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000013526 transfer learning Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformation in the plane of the image
- G06T3/40—Scaling the whole image or part thereof
- G06T3/4007—Interpolation-based scaling, e.g. bilinear interpolation
-
- G06T5/70—
-
- G06T5/90—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20172—Image enhancement details
- G06T2207/20192—Edge enhancement; Edge preservation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30248—Vehicle exterior or interior
Abstract
The invention discloses an illegal hazardous chemical substance transport vehicle identification method based on a convolutional neural network, which comprises the following steps of: s100, acquiring training data and carrying out data standardization processing; s200, carrying out convolution operation by utilizing a convolution neural network calculation model to obtain a trained calculation model and a final labeled image library; s300, a local dangerous chemical vehicle database and data to be identified are obtained, the data to be identified are calculated and identified, the data to be identified are compared with a final labeled image database for judgment, the characteristics of dangerous chemical vehicle or suspected dangerous chemical vehicle are output, the data and the local dangerous chemical vehicle database are judged, and standard dangerous chemical transport vehicles and illegal dangerous chemical transport vehicles are distinguished. The invention also discloses an illegal hazardous chemical substance transport vehicle identification system based on the convolutional neural network, and the standard hazardous chemical substance transport vehicle and the illegal hazardous chemical substance transport vehicle in the traffic picture stream can be identified by utilizing the image identification and the convolutional neural network calculation model, so that the time consumption is low, and the labor cost is reduced.
Description
Technical Field
The invention relates to the technical field of intelligent identification of dangerous chemical substance flows, in particular to an illegal dangerous chemical substance transport vehicle identification method and system based on a convolutional neural network. Specifically, G06Q10/08 belongs to the IPC classification.
Background
With the rapid development of economy in China, the traffic volume of roads and logistics is increasing continuously, and the safety and smoothness of traffic become the central importance, especially for the storage and logistics of dangerous chemicals. Due to the characteristics of flammability, explosiveness, easy poisoning, easy pollution and the like, accidents occur in the transportation process, and huge loss is caused. For the management and control transportation of hazardous chemical substances, the coordinated operation of a plurality of departments is required, the departments need real-time information intercommunication, and specific information needs to be displayed for different departments under specific conditions according to the needs so as to improve the utilization efficiency of the information.
Prior art 1: patent application No. CN206610441U discloses an intelligent dangerous chemical substance vehicle regional access monitoring system, including a video/radio frequency identification device installed at the access of a control region, a positioning module and a vehicle identity information module installed on a dangerous chemical substance logistics vehicle, and a dangerous chemical substance regional access control facility installed at a dangerous chemical substance control center. According to the technical scheme, the control of the logistics of the hazardous chemical substances of specific types in a specific area can be realized, but in addition to standard special transportation vehicles for the hazardous chemical substances, some non-qualified vehicles can also transport the related hazardous chemical substances on the daily road surface, the control of the transportation of the hazardous chemical substances by the non-qualified vehicles cannot be realized, and the control is not suitable for the comprehensive supervision of the transportation vehicles for the hazardous chemical substances in the traffic roads.
Prior art 2: the patent application number is CN110689306A discloses a dangerous chemical substance road transportation management system based on two-dimensional codes and an operation method thereof, the system comprises a two-dimensional code decoding device, a two-dimensional code label arranged on a dangerous chemical substance transportation vehicle, a data management server, a dangerous chemical substance road transportation information acquisition device, a dangerous chemical substance basic information database server and a web server, wherein the dangerous chemical substance road transportation information acquisition device, the dangerous chemical substance basic information database server and the web server are respectively connected with the data management server through a network. According to the technical scheme, the two-dimensional code label on the dangerous chemical transport vehicle is identified, so that the illegal dangerous chemical transport vehicle cannot be identified.
In summary, the prior art supervises qualified legal transportation vehicles for hazardous chemical substances, and besides standard transportation vehicles special for hazardous chemical substances, some non-qualified vehicles are also transporting related hazardous chemical substances in daily roads, and identification of such vehicles is very difficult. How to find out the information and the track of dangerous chemicals or suspected dangerous chemicals in huge traffic flow is a practical problem. If all roads are monitored manually, the method is long in time consumption and high in labor cost, and cannot be realized.
Disclosure of Invention
The invention aims to provide an illegal hazardous chemical substance transport vehicle identification method and system based on a convolutional neural network, which can extract information and tracks of hazardous chemical substances or suspected hazardous chemical substances from traffic picture streams, identify standard hazardous chemical substance transport vehicles and illegal hazardous chemical substance transport vehicles and reduce labor cost.
In order to achieve the purpose, the invention adopts the following technical scheme:
the illegal hazardous chemical substance transport vehicle identification method and system based on the convolutional neural network comprise the following steps:
s100, acquiring training data, and then performing data standardization processing on the training data, wherein the training data is an initial labeled image library, and the initial labeled image library comprises classified dangerous chemical substance pictures, dangerous chemical substance icons and standard dangerous chemical substance transport vehicle illustrations;
s200, carrying out convolution operation on the processed training data by utilizing a convolution neural network calculation model, and learning, training and correcting to obtain a trained calculation model and a final labeled image library;
s300, a local dangerous chemical vehicle database and data to be identified are obtained, the data to be identified are calculated and identified by using a trained calculation model, the data to be identified are compared and judged with a final labeled image library, the characteristics of dangerous chemical vehicles or suspected dangerous chemical vehicles are output, the characteristics of the dangerous chemical vehicles or the suspected dangerous chemical vehicles are judged with the local dangerous chemical vehicle database, and standard dangerous chemical transport vehicles and illegal dangerous chemical transport vehicles are distinguished.
Wherein step S100 further comprises:
s110, image gray level processing: performing Gaussian blur on an image of training data to reduce image noise, and then performing weighted gray processing on three channels of R, G and B of the image;
s120, image edge analysis: detecting edges in the gray images by using a canny algorithm, and outputting image information;
s130, image cutting: based on the image edge analysis result, cutting the part which does not influence the content in the image, and dividing the part into a plurality of different new images so as to improve the recognition rate;
s140, vectorizing the image; vectorizing the image, converting the image into a planar two-dimensional array, wherein the gray value information of each pixel point represents the information of one feature in the vector;
s150, image standardization: the resolution of the image information is unified, and the normalization processing is performed.
The step S150 further includes:
s151, defining the resolution of the input image to be 200 x 200;
s152, rewriting the resolution of the image to 200 x 200 by adopting a nearest neighbor interpolation method;
s153, carrying out normalization processing on the vector group of the image, and placing the obtained result in the original position to obtain a normalized vector group.
Step S200 further includes:
s210, convolution operation and prediction: inputting the processed training data into a convolutional neural network calculation model, wherein the convolutional neural network calculation model comprises two convolutional networks A and B, the convolutional network A is a normalized vector convolutional network processed by Gaussian blur-graying, the convolutional network B is a normalized vector convolutional network of Gaussian blur-RGB full channels, a prediction result is obtained through calculation, the prediction result comprises A classification (A class), A score (A score), B classification (B class) and B score (B class), and the initial weights of the prediction result on classification are respectively specified as
And
，
；
s220, excitation is carried out after verification and judgment: respectively verifying and judging the A classification and the B classification in each prediction result and the classification marked by the corresponding training data, and if the A classification and the B classification are both judged to be correct or both judged to be wrong, keeping the weight unchanged, namely
，
(ii) a If only one of the A-class and the B-class is correctly determined, the positive excitation is correctly determined, the negative excitation is incorrectly determined, i.e. if the A-class is correctly determined and the B-class is incorrectly determined,
，
when the B classification decision is correct and the A classification decision is incorrect,
，
(ii) a Wherein the content of the first and second substances,
in order to obtain the excitation rate of the posterior part,
，
for the weights that influence the a classification after excitation,
weights for the excited influence B classes;
s230, comparing the classification A and the classification B and corresponding weights, and outputting a classification result R and a grading result P: when the a classification and the B classification are the same,
，
(ii) a When A classification and B classification are different
When the temperature of the water is higher than the set temperature,
，
(ii) a When A classification and B classification are different
When the temperature of the water is higher than the set temperature,
，
(ii) a When A classification and B classification are different
When the image is marked, the corresponding image enters a manual marking library;
s240, setting a grading threshold value
When is coming into contact with
Then, inputting the corresponding initial marked image into a marked image library, and marking according to prediction classification;when in use
Then, the corresponding initial annotation image enters a manual annotation library;
s250, manual labeling: manually labeling the images in the manual labeling library, outputting the images to a corrected labeling image library, and weighting corresponding to the judgment result
、
Correcting;
s260, relearning: and inputting the corrected marked image library after artificial marking into the convolutional neural network calculation model again for prediction calculation, and repeatedly training and correcting the convolutional neural network calculation model to obtain a trained calculation model and a final marked image library.
The local dangerous chemical vehicle database in the step S300 provides registered standard dangerous chemical transport vehicle information for the local transportation administrative department, the data to be identified is a batch traffic picture stream, and the batch traffic picture stream includes vehicle body dangerous chemical icons, cargo dangerous chemical pictures, vehicle traveling tracks and vehicle information.
Step S300 further includes:
s310, acquiring an automobile body hazardous chemical substance icon (a sealed automobile), a cargo hazardous chemical substance picture (a non-sealed automobile), a vehicle running track and vehicle information in the data to be identified;
s320, inputting the vehicle body hazardous chemical substance icon and the cargo hazardous chemical substance picture into a trained convolutional neural network calculation model for calculation, comparing and judging with a final labeled image library, and outputting hazardous chemical substance vehicle characteristics or suspected hazardous chemical substance vehicle characteristics;
s330, comparing the characteristics of the dangerous chemical substance vehicles or the characteristics of the suspected dangerous chemical substance vehicles with the local dangerous chemical substance vehicle database, if the characteristics are matched, the vehicles are transported by legal dangerous chemical substance vehicles, and if more than one of the characteristics are not matched, the vehicles are transported by illegal dangerous chemical substance vehicles.
Step S330 further includes:
s331, identifying whether the license plate is an effective license plate through an image, comparing the license plate with a local hazardous chemical substance vehicle database, and correspondingly judging as a license-plate-free hazardous chemical substance transport vehicle or a fake plate hazardous chemical substance transport vehicle if no license plate exists on the vehicle or the license plate is an ineffective license plate;
s332, identifying the type of goods on the non-sealed vehicle through the image, comparing the operation type with the local vehicle information, and if the operation types are not consistent, judging that no qualification certificate dangerous chemical transport vehicle exists;
and S333, if the vehicle track data does not belong to the local dangerous chemical vehicle database, judging that the vehicle is a cross-domain dangerous chemical transport vehicle/illegal destination dangerous chemical transport vehicle.
The invention also provides an illegal hazardous chemical substance transport vehicle identification system based on the convolutional neural network, which can realize the method and comprises the following steps:
a first data input processing unit: the image processing system is used for acquiring training data and carrying out standardized processing on the training data, wherein the training data is an initial labeling image library;
a convolution operation unit: the first data input processing unit is connected with the first data input processing unit and is used for performing convolution operation on the processed training data and obtaining a final labeled image library after learning training;
a second data input processing unit: the convolution operation unit is connected with the vehicle identification device and used for acquiring data to be identified and inputting the data to be identified into the trained convolution operation unit to acquire the characteristics of the dangerous chemical vehicle or the suspected dangerous chemical vehicle;
a determination output unit: and the second data input processing unit is connected with the first data input processing unit and is used for acquiring a local dangerous chemical vehicle database, comparing the dangerous chemical vehicle characteristics or suspected dangerous chemical vehicle characteristics with the final labeled image database and outputting a judgment result.
The invention also provides a computer-readable storage medium, on which a computer program is stored, which when executed implements the method described above.
The present invention also provides an electronic device, comprising:
a processor; and
a memory for storing executable instructions of the processor;
wherein the processor is configured to perform the method described above via execution of the executable instructions.
In conclusion, the beneficial technical effects of the invention are as follows:
(1) according to the invention, the traffic picture stream is identified in batch by utilizing the image identification and convolution neural network calculation model, information and track of dangerous chemicals or suspected dangerous chemicals are extracted, and standard dangerous chemical transport vehicles and illegal dangerous chemical transport vehicles can be identified.
(2) The convolution result and the labeling information of the labeled picture are used as a training set of the convolution neural network calculation model, the model is trained through a gradient lifting algorithm, the image attribute labeling precision can be effectively improved, the model is adjusted through the modes of reserving training data, cross validation, manual labeling correction and the like, the recognition accuracy of the calculation model to the training set reaches an expected result, and the prediction accuracy of the model to the traffic picture stream is finally improved.
(3) Meanwhile, the illegal dangerous chemical substance transport vehicles can be subdivided into cross-domain dangerous chemical substance transport vehicles/illegal destination dangerous chemical substance transport vehicles, the output results are visual and easy to understand, and the later tracking and management of traffic supervision personnel are facilitated.
(4) When the convolutional neural network calculation model is used for learning and training, the model can store the form and parameters of the model and can be transplanted to other system environments at any time, and the flexibility and the portability of the model are improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
Fig. 1 is a schematic workflow diagram of an illegal hazardous chemical substance transportation vehicle identification method based on a convolutional neural network in embodiment 1;
FIG. 2 is an example of the results of graying, Gaussian blur, and edge analysis of the explosives icon in example 1;
FIG. 3 is an example of the result of image segmentation in example 1;
fig. 4 is an example of an image vectorization array obtained after vectorization processing of an image in embodiment 1;
FIG. 5 is a flowchart showing the structural operation of the convolutional neural network computational model in example 1;
fig. 6 is a schematic structural diagram of an illegal hazardous chemical substance transportation vehicle identification system based on a convolutional neural network in embodiment 2;
fig. 7 is a schematic structural diagram of an electronic device in embodiment 3.
Reference numerals: 401. a first data input processing unit; 402. a convolution operation unit; 403. a second data input processing unit; 404. a determination output unit; 501. a processor; 502. a memory; 503. a communication interface.
Detailed Description
The technical solutions of the present invention will be described clearly and completely with reference to the accompanying drawings, and it should be understood that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, and are only for convenience of description and simplicity of description, but do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus, should not be construed as limiting the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
In the description of the present invention, it should be noted that, unless otherwise explicitly specified or limited, the terms "mounted," "connected," and "connected" are to be construed broadly, e.g., as meaning either a fixed connection, a removable connection, or an integral connection; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood by those skilled in the art according to specific situations.
The technical terms referred to in the present document are explained first below:
standard hazardous chemical substance transport vehicle: generally, the vehicle is qualified and transports dangerous chemical substances in a legal driving time and a driving road.
Vehicle of illegal dangerous chemical transport vehicle: the vehicle is characterized by having the behaviors of unlicensed, fake-licensed, cross-domain, unqualified and illegal destination transportation of dangerous chemicals.
Gaussian Blur (Gaussian Blur): also known as gaussian smoothing, is generally used to reduce image noise and detail level, and is also used in a preprocessing stage in a computer vision algorithm to enhance the image effect of images in different scale sizes.
R, G, B weight: the RGB color scheme is a color standard in the industry, and various colors are obtained by changing three color channels of red (R), green (G) and blue (B) and superimposing them on each other. In the weighted gradation processing, gradation value calculation is performed by weighted sampling of three color channel values of RGB.
canny algorithm: the method is a multi-stage edge detection algorithm, can remarkably reduce the data scale of an image under the condition of keeping the original image attribute, and is divided into the following 5 steps: 1) applying gaussian filtering to smooth the image with the aim of removing noise; 2) finding intensity gradients (intensity gradients) of the image; 3) applying a non-maximum suppression (non-maximum suppression) technique to eliminate edge false detection (which is not originally detected but detected); 4) applying a dual threshold approach to determine possible (potential) boundaries; 5) the boundaries are tracked using a hysteresis technique.
Nearest neighbor interpolation: and assigning the gray value of the nearest pixel of the original pixel point in the transformed image to the original pixel point.
A Convolutional Neural Network (CNN) is a feed-forward Neural Network whose artificial neurons can respond to a portion of the coverage of surrounding cells, and performs well for large image processing. The convolutional neural network consists of one or more convolutional layers and a top fully connected layer (corresponding to the classical neural network), and also includes associated weights and pooling layers (pooling layers).
And (3) rolling layers: each convolution layer in the convolutional neural network consists of a plurality of convolution units, and the parameters of each convolution unit are optimized through a back propagation algorithm. The convolution operation aims to extract different input features, the convolution layer at the first layer can only extract some low-level features such as edges, lines, angles and other levels, and more layers of networks can iteratively extract more complex features from the low-level features.
Gradient Boosting (Gradient Boosting) algorithm: the principle of the method is that a newly added weak classifier is trained according to negative Gradient information of a current model loss function, and then the trained weak classifier is combined into an existing model in an accumulated form. Specific applications can refer to image attribute labeling based on extreme gradient lifting tree algorithm published in stew and bin, cudweifer, Wu, billow, claw, and brough No. 4 months in 2019, which article indicates that: an image attribute annotation model based on an eXtreme gradient boosting tree (XGBoost) algorithm is used to improve the annotation performance: extracting the characteristics of Local Binary Patterns (LBP), gray texture space envelope characteristics (Gist), Scale Invariant Feature Transform (SIFT), Visual Geometry Group (VGG) and the like of the image so as to accurately depict the visual content of the image; deep semantics contained in the image attributes are deeply excavated, and a brand-new hierarchical attribute representation system is constructed so as to be close to objective cognition of human beings; and designing a transfer learning strategy and reasonably combining a classification model to further improve the labeling performance. Tests show that Gist characteristics can truly depict image visual contents, the marking accuracy is improved by 8.69% compared with the optimal index before migration learning after basic migration learning is executed, and the marking accuracy is improved by 17.55% compared with the optimal index before basic migration learning by reasonably combining classification models after mixed type migration learning is executed. The model effectively improves the image attribute labeling precision.
Cross-validation (Cross-validation): the method is mainly used for modeling applications, such as PCR and PLS regression modeling. In a given modeling sample, most samples are taken out to build a model, a small part of samples are reserved to be forecasted by the just built model, forecasting errors of the small part of samples are solved, and the sum of squares of the forecasting errors is recorded. The purpose of cross-validation is to obtain a reliable and stable model.
Example 1
Referring to fig. 1, the illegal hazardous chemical substance transportation vehicle identification method based on the convolutional neural network disclosed by the invention comprises the following steps:
s100, training data are obtained, and then data standardization processing is carried out on the training data, wherein the training data are an initial labeling image library, and the initial labeling image library comprises classified dangerous chemical substance pictures, dangerous chemical substance icons and standard dangerous chemical substance transport vehicle illustrations.
Wherein, step S100 further comprises:
s110, image gray level processing: the image of the training data is subjected to Gaussian blur to reduce image noise, and then the R, G and B channels of the image are subjected to weighted gray scale processing, wherein in the embodiment, the weights of R, G and B are respectively selected from 0.299,0.587 and 0.114.
S120, image edge analysis: and detecting edges in the gray images by using a canny algorithm, and outputting image information. An example of the results of graying, gaussian blur, and edge analysis of explosives icons is shown in fig. 2.
S130, image cutting: based on the image edge analysis result, the part which does not influence the content in the image is cut and divided into a plurality of different new images, so that the recognition rate is improved, and the effective content of the image can be minimized. An example of the result of the image segmentation is shown with reference to fig. 3.
S140, vectorizing the image; vectorizing the image, converting the image into a planar two-dimensional array, wherein the gray value information of each pixel point represents the information of one feature in the vector. An example of an image vectorization array obtained through vectorization processing is shown in fig. 4.
S150, image standardization: the resolution of the image information is unified, and the normalization processing is performed.
The image normalization further comprises the steps of:
s151, defining the resolution of the input image to be 200 x 200;
s152, rewriting the resolution of the image to 200 x 200 by adopting a nearest neighbor interpolation method;
and S153, carrying out normalization processing on the vector group of the image, uniformly removing 255.0, and placing the obtained result in the original position to obtain a normalized vector group.
S200, carrying out convolution operation on the processed training data by utilizing a convolution neural network calculation model, and learning, training and correcting to obtain a trained calculation model and a final labeled image library.
Wherein, step S200 further comprises:
s210, convolution operation and prediction: inputting the processed training data into a convolutional neural network computational model, referring to fig. 5, the convolutional neural network computational model includes two initial convolutional layers a and B, which are convolutional layers + pooling layers respectively, and the combination of the convolutional layers + pooling layers may occur for a plurality of times, and the occurrence number is set according to the needs of the model, and is set to two times in this embodiment. The purpose of the convolution operation is to extract different features of the input, the first tier volumeThe lamination layer can only extract some low-level features such as edges, lines, angles and other levels, and more layers of model structures can iteratively extract more complex features from the low-level features so as to meet the actual calculation requirement. The initial convolution layer A is a normalized vector convolution network of Gaussian blur-graying processing, the convolution network B is a normalized vector convolution network of Gaussian blur-RGB full channels, then convolution operation is carried out to obtain a prediction result, the prediction result comprises A classification (A class), A score (A score), B classification (B class) and B score (B class), and initial weights of the prediction result on classification are respectively specified as
And
，
；
s220, excitation is carried out after verification and judgment: respectively verifying and judging the A classification and the B classification in each prediction result and the classification marked by the corresponding training data, and if the A classification and the B classification are both judged to be correct or both judged to be wrong, keeping the weight unchanged, namely
，
(ii) a If only one of the A-class and the B-class is correctly determined, the positive excitation is correctly determined, the negative excitation is incorrectly determined, i.e. if the A-class is correctly determined and the B-class is incorrectly determined,
，
when the B classification decision is correct and the A classification decision is incorrect,
，
(ii) a Wherein the content of the first and second substances,
in order to obtain the excitation rate of the posterior part,
，
for the weights that influence the a classification after excitation,
weights that influence the B classification after excitation. The prediction results after excitation are shown in table 1 below:
TABLE 1 prediction result determination and weight variation correspondence table
S230, comparing the classification A and the classification B and corresponding weights, and outputting a classification result R and a grading result P: when the a classification and the B classification are the same,
，
(ii) a When A classification and B classification are different
When the temperature of the water is higher than the set temperature,
，
(ii) a When A classification and B classification are different
When the temperature of the water is higher than the set temperature,
，
(ii) a When A classification and B classification are different
And when the corresponding image enters the manual annotation library. The corresponding output results are shown in table 2 below:
TABLE 2 comparison result and output result correspondence table
S240, setting a grading threshold value
In the present embodiment, the first and second electrodes,
is set to 0 when
Then, inputting the corresponding initial marked image into a marked image library, and marking according to prediction classification; when in use
And then, the corresponding initial annotation image enters a manual annotation library.
S250, manual labeling: manually labeling the images in the manual labeling library, outputting the images to a corrected labeling image library, and weighting corresponding to the judgment result
、
And (6) correcting.
Wherein, the pictures in the revised labeled image library are stored in the forms of label classification and label two-dimensional vector group, for example: label classification 1003_ explicit.jpg, label vector 1003_ explicit.vec.
S260, relearning: and inputting the corrected marked image library after artificial marking into the convolutional neural network calculation model again for prediction calculation, and repeatedly training and correcting the convolutional neural network calculation model to obtain a trained calculation model and a final marked image library.
The convolution result and the labeling information of the labeled picture are used as a training set of the convolution neural network calculation model, the model is trained through a gradient lifting algorithm, the image attribute labeling precision can be effectively improved, and the model is adjusted through the modes of reserving training data, cross validation, manual labeling correction and the like, so that the recognition accuracy of the calculation model to the training set reaches an expected result.
S300, a local dangerous chemical vehicle database and data to be identified are obtained, the trained calculation model is used for calculating and identifying the data to be identified, the data to be identified is compared with a final labeled image library for judgment, and the characteristics of dangerous chemical vehicles or suspected dangerous chemical vehicles are output. And judging the characteristics of the dangerous chemical substance vehicles or the characteristics of the suspected dangerous chemical substance vehicles and a local dangerous chemical substance vehicle database, and distinguishing standard dangerous chemical substance transport vehicles and illegal dangerous chemical substance transport vehicles.
The local dangerous chemical vehicle database provides registered standard dangerous chemical transport vehicle information for local traffic authorities, wherein the registered standard dangerous chemical transport vehicle information comprises but is not limited to information such as license plates, dangerous chemical species, vehicle body registration colors, vehicle body length and width, the data to be identified is a batch traffic picture stream, and the batch traffic picture stream comprises vehicle body dangerous chemical icons (sealed vehicles), cargo dangerous chemical pictures (non-sealed vehicles), vehicle running tracks and vehicle information.
Step S300 further includes:
s310, vehicle body hazardous chemical substance icons (sealed vehicles), cargo hazardous chemical substance pictures (non-sealed vehicles), vehicle running tracks and vehicle information in the data to be identified are obtained, wherein the vehicle information comprises characteristics of license plates, vehicle body colors, vehicle body length and width and the like.
Aiming at the sealed vehicle, identifying, calculating and predicting dangerous chemical icons attached or printed on the vehicle body; for a non-sealed vehicle, identification, calculation and prediction of images of dangerous chemical substances of cargos on the vehicle are required.
And S320, inputting the vehicle body dangerous chemical icon and the cargo dangerous chemical picture into the trained convolutional neural network calculation model for calculation and prediction, comparing and judging the vehicle body dangerous chemical icon and the cargo dangerous chemical picture with a final labeled image library, and outputting dangerous chemical vehicle characteristics or suspected dangerous chemical vehicle characteristics. The dangerous chemical substance vehicle characteristics or suspected dangerous chemical substance vehicle characteristics comprise dangerous chemical substance classification, vehicle license plate information, vehicle driving track data and vehicle information.
The classification of dangerous chemicals includes 8 categories of explosives, compressed gas and liquefied gas, flammable liquid, etc. Example of the output prediction results: and [ 'applied', 34.73], which indicates that the dangerous chemical corresponding to the image is classified as an explosive.
S330, comparing the characteristics of the dangerous chemical substance vehicles or the characteristics of the suspected dangerous chemical substance vehicles with the local dangerous chemical substance vehicle database, if the characteristics are matched, the vehicles are transported by legal dangerous chemical substance vehicles, and if more than one of the characteristics are not matched, the vehicles are transported by illegal dangerous chemical substance vehicles.
Wherein, the judgment of the illegal dangerous chemical vehicle transportation vehicle further comprises the following steps:
s331, identifying whether the license plate is a valid license plate through an image, comparing the license plate with a local hazardous chemical substance vehicle database, and judging as a license-free hazardous chemical substance transport vehicle if no license plate exists on the vehicle; and if the license plate is an invalid license plate, judging that the vehicle is a fake plate dangerous chemical transport vehicle.
The specific judgment process of the fake-licensed dangerous chemical substance transport vehicle is to judge whether the license plate, the color of the vehicle body and the length and width of the vehicle body are consistent with the registered information, and if the license plate, the color of the vehicle body and the length and width of the vehicle body are inconsistent, the fake-licensed dangerous chemical substance transport vehicle is determined.
S332, identifying the type of the goods on the non-sealed vehicle through the image, comparing the operation type with the local vehicle information, and if the operation types are not consistent, judging that no qualification certificate dangerous chemical transport vehicle exists.
For example, some small transport vehicles illegally transport goods such as gas tanks under the condition that related certificates issued by a local traffic management part are not obtained, and the condition is judged to be a dangerous chemical transport vehicle without qualification certificate.
And S333, if the vehicle track data does not belong to the local dangerous chemical vehicle database, judging that the vehicle is a cross-domain dangerous chemical transport vehicle/illegal destination dangerous chemical transport vehicle.
The working principle and the beneficial effects of the embodiment are as follows:
in this embodiment, the convolution result and the labeling information of the labeled picture are used as a training set of the convolutional neural network computational model, the model is trained through a gradient lifting algorithm, and the model is adjusted through ways of reserving training data, cross validation, manual labeling correction and the like, so that the recognition accuracy of the computational model to the training set reaches an expected result, and the prediction accuracy of the model to the traffic picture stream is finally improved. The traffic picture stream is identified in batches by utilizing an image identification and convolution neural network calculation model, information and tracks of dangerous chemicals or suspected dangerous chemicals are extracted, standard dangerous chemical transport vehicles and illegal dangerous chemical transport vehicles are identified, the illegal dangerous chemical transport vehicles can be further subdivided, output results are visual and easy to understand, and later-stage tracking and management of traffic supervisors are facilitated. Compared with manual identification and manual supervision, the method consumes less time, and greatly reduces the labor cost.
Example 2
Referring to fig. 6, the illegal hazardous chemical substance vehicle identification system based on the convolutional neural network includes a first data input processing unit 401, a convolutional operation unit 402, a second data input processing unit 403, and a determination output unit 404, where the convolutional operation unit 402 is connected to the first data input processing unit 401, the second data input processing unit 403 is connected to the convolutional operation unit 402, and the determination output unit 404 is connected to the second data input processing unit 403.
The first data input processing unit 401 is configured to acquire training data and perform normalization processing on the training data, where the training data is an initial labeled image library.
The convolution operation unit 402 includes a convolution neural network computation model, and is configured to perform convolution operation on the processed training data, and obtain a final labeled image library after learning and training. When the convolutional neural network calculation model is used for learning and training, the model can store the form and parameters of the model and can be transplanted to other system environments at any time, and the flexibility and the portability of the model are improved.
The second data input processing unit 403 is configured to obtain data to be identified, input the data to be identified into the trained convolution operation unit 402, and obtain a hazardous chemical substance vehicle characteristic or a suspected hazardous chemical substance vehicle characteristic.
The determination output unit 404 is configured to obtain a local hazardous chemical substance vehicle database, compare the hazardous chemical substance vehicle characteristics or suspected hazardous chemical substance vehicle characteristics with the final labeled image library, and output a determination result.
The system can implement the corresponding method in the foregoing method embodiment, and the specific implementation process thereof can refer to the foregoing method embodiment, which is not described herein again.
Example 3
Referring to fig. 7, an embodiment of an electronic device provided in the present invention includes: a processor 501, and a memory 502 for storing executable instructions for the processor 501. Optionally, the method may further include: a communication interface 503 for communicating with other devices.
The processor 501 is configured to execute the method corresponding to the foregoing method embodiment by executing the executable instruction, and the specific implementation process of the method may refer to the foregoing method embodiment, which is not described herein again.
Example 4
The present invention further provides a computer-readable storage medium, on which a computer program (or called computer-executable instructions) is stored, where when the computer program is executed, the method corresponding to the foregoing method embodiment can be implemented, and a specific implementation process of the method may refer to the foregoing method embodiment, which is not described herein again.
The computer readable storage medium is a magnetic disk, an optical disk, a read-only memory or a random access memory.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.
Claims (9)
1. The illegal hazardous chemical substance transport vehicle identification method based on the convolutional neural network is characterized by comprising the following steps of:
s100, acquiring training data, and then performing data standardization processing on the training data, wherein the training data is an initial labeled image library, and the initial labeled image library comprises classified dangerous chemical substance pictures, dangerous chemical substance icons and standard dangerous chemical substance transport vehicle illustrations;
s200, carrying out convolution operation on the processed training data by utilizing a convolution neural network calculation model, and learning, training and correcting to obtain a trained calculation model and a final labeled image library;
step S200 further includes:
s210, convolution operation and prediction: inputting the processed training data into a convolutional neural network computation model, wherein the convolutional neural network computation model comprises two convolutional networks A and B, the convolutional network A is a normalized vector convolutional network processed by Gaussian blur-graying, the convolutional network B is a normalized vector convolutional network of Gaussian blur-RGB full channels, a prediction result is obtained through computation, the prediction result comprises A classification, A scoring, B classification and B scoring, and the A classification, the A scoring, the B classification and the B scoring are respectively used
、
、
、
Expressing and defining initial weights of the prediction results on the classification respectively as
And
，
；
s220, excitation is carried out after verification and judgment: respectively verifying and judging the A classification and the B classification in each prediction result and the classification marked by the corresponding training data, and if the A classification and the B classification are both judged to be correct or both judged to be wrong, keeping the weight unchanged, namely
，
(ii) a If only one of the A-class and the B-class is correctly determined, the positive excitation is correctly determined, the negative excitation is incorrectly determined, i.e. if the A-class is correctly determined and the B-class is incorrectly determined,
，
when the B classification decision is correct and the A classification decision is incorrect,
，
(ii) a Wherein the content of the first and second substances,
in order to obtain the excitation rate of the posterior part,
，
for the weights that influence the a classification after excitation,
weights for the excited influence B classes;
s230, comparing the classification A and the classification B and corresponding weights, and outputting a classification result R and a grading result P: when the a classification and the B classification are the same,
，
(ii) a When A classification and B classification are different
When the temperature of the water is higher than the set temperature,
，
(ii) a When A classification and B classification are different
When the temperature of the water is higher than the set temperature,
，
(ii) a When A classification and B classification are different
When the image is marked, the corresponding image enters a manual marking library;
s240, setting a grading threshold value
When is coming into contact with
Then, inputting the corresponding initial marked image into a marked image library, and marking according to prediction classification; when in use
Then, the corresponding initial annotation image enters a manual annotation library;
s250, manual labeling: manually labeling the images in the manual labeling library, outputting the images to a corrected labeling image library, and weighting corresponding to the judgment result
、
Correcting;
s260, relearning: based on the corrected labeled image library after artificial labeling, inputting the corrected labeled image library into the convolutional neural network calculation model again for prediction calculation, and repeatedly training and correcting the convolutional neural network calculation model to obtain a trained calculation model and a final labeled image library;
s300, a local dangerous chemical vehicle database and data to be identified are obtained, the data to be identified are calculated and identified by using a trained calculation model, the data to be identified are compared and judged with a final labeled image library, the characteristics of dangerous chemical vehicles or suspected dangerous chemical vehicles are output, the characteristics of the dangerous chemical vehicles or the suspected dangerous chemical vehicles are judged with the local dangerous chemical vehicle database, and standard dangerous chemical transport vehicles and illegal dangerous chemical transport vehicles are distinguished.
2. The identification method for illegal hazardous chemical substance transportation vehicles based on convolutional neural network as claimed in claim 1, wherein step S100 further comprises:
s110, image gray level processing: performing Gaussian blur on an image of training data to reduce image noise, and then performing weighted gray processing on three channels of R, G and B of the image;
s120, image edge analysis: detecting edges in the gray images by using a canny algorithm, and outputting image information;
s130, image cutting: based on the image edge analysis result, cutting the part which does not influence the content in the image, and dividing the part into a plurality of different new images so as to improve the recognition rate;
s140, vectorizing the image; vectorizing the image, converting the image into a planar two-dimensional array, wherein the gray value information of each pixel point represents the information of one feature in the vector;
s150, image standardization: the resolution of the image information is unified, and the normalization processing is performed.
3. The identification method for illegal hazardous chemical substance transportation vehicles based on convolutional neural network as claimed in claim 2, wherein said step S150 further comprises:
s151, defining the resolution of the input image to be 200 x 200;
s152, rewriting the resolution of the image to 200 x 200 by adopting a nearest neighbor interpolation method;
s153, carrying out normalization processing on the vector group of the image, and placing the obtained result in the original position to obtain a normalized vector group.
4. The method for identifying illegal hazardous chemical substance transport vehicles based on convolutional neural network as claimed in claim 1, wherein the local hazardous chemical substance database in step S300 provides registered standard hazardous chemical substance transport vehicle information for local transportation authorities, and the data to be identified is a batch traffic picture stream, and the batch traffic picture stream comprises vehicle body hazardous chemical substance icons, cargo hazardous chemical substance pictures, vehicle driving tracks and vehicle information.
5. The identification method for illegal hazardous chemical substance transportation vehicles based on convolutional neural network as claimed in claim 4, wherein step S300 further comprises:
s310, acquiring an automobile body hazardous chemical substance icon, a cargo hazardous chemical substance picture, an automobile driving track and automobile information in the data to be identified;
s320, inputting the vehicle body hazardous chemical substance icon and the cargo hazardous chemical substance picture into a trained convolutional neural network calculation model for calculation, comparing and judging with a final labeled image library, and outputting hazardous chemical substance vehicle characteristics or suspected hazardous chemical substance vehicle characteristics;
s330, comparing the characteristics of the dangerous chemical substance vehicles or the characteristics of the suspected dangerous chemical substance vehicles with the local dangerous chemical substance vehicle database, if the characteristics are matched, the vehicles are transported by legal dangerous chemical substance vehicles, and if more than one of the characteristics are not matched, the vehicles are transported by illegal dangerous chemical substance vehicles.
6. The identification method for illegal hazardous chemical substance transportation vehicles based on convolutional neural network as claimed in claim 5, wherein step S330 further comprises:
s331, identifying whether the license plate is an effective license plate through an image, comparing the license plate with a local hazardous chemical substance vehicle database, and correspondingly judging as a license-plate-free hazardous chemical substance transport vehicle or a fake plate hazardous chemical substance transport vehicle if no license plate exists on the vehicle or the license plate is an ineffective license plate;
s332, identifying the type of goods on the non-sealed vehicle through the image, comparing the operation type with the local vehicle information, and if the operation types are not consistent, judging that no qualification certificate dangerous chemical transport vehicle exists;
and S333, if the vehicle track data does not belong to the local dangerous chemical vehicle database, judging that the vehicle is a cross-domain dangerous chemical transport vehicle/illegal destination dangerous chemical transport vehicle.
7. The system for identifying the illegal hazardous chemical substance transport vehicle based on the convolutional neural network is used for realizing the method of any one of the claims 1-6, and is characterized by comprising the following steps:
a first data input processing unit (401): the image processing system is used for acquiring training data and carrying out standardized processing on the training data, wherein the training data is an initial labeling image library;
convolution operation unit (402): the first data input processing unit (401) is connected with the first data input processing unit and is used for performing convolution operation on the processed training data and obtaining a final labeled image library after learning training;
a second data input processing unit (403): the convolution operation unit (402) is connected and used for acquiring data to be identified, inputting the data to be identified into the trained convolution operation unit (402) and acquiring the dangerous chemical substance vehicle characteristics or the suspected dangerous chemical substance vehicle characteristics;
determination output unit (404): and the second data input processing unit (403) is connected and used for acquiring a local dangerous chemical vehicle database, comparing the dangerous chemical vehicle characteristics or suspected dangerous chemical vehicle characteristics with the final labeled image library and outputting a judgment result.
8. A computer-readable storage medium, on which a computer program is stored, which, when executed, implements the method of any of claims 1 to 6.
9. An electronic device, comprising:
a processor (501); and
a memory (502) for storing executable instructions of the processor (501);
wherein the processor (501) is configured to perform the method of any of claims 1-6 via execution of the executable instructions.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111035738.8A CN113469158B (en) | 2021-09-06 | 2021-09-06 | Method and system for identifying illegal hazardous chemical substance transport vehicle based on convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111035738.8A CN113469158B (en) | 2021-09-06 | 2021-09-06 | Method and system for identifying illegal hazardous chemical substance transport vehicle based on convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113469158A CN113469158A (en) | 2021-10-01 |
| CN113469158B true CN113469158B (en) | 2021-11-19 |
Family
ID=77867483
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111035738.8A Active CN113469158B (en) | 2021-09-06 | 2021-09-06 | Method and system for identifying illegal hazardous chemical substance transport vehicle based on convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113469158B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114155495B (en) * | 2022-02-10 | 2022-05-06 | 西南交通大学 | Safety monitoring method, device, equipment and medium for vehicle operation in sea-crossing bridge |
| CN114627653B (en) * | 2022-05-12 | 2022-08-02 | 浙江电马云车科技有限公司 | 5G intelligent barrier gate management system based on binocular recognition |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107886731A (en) * | 2017-11-03 | 2018-04-06 | 武汉元鼎创天信息科技有限公司 | A kind of illegal operation Vehicular intelligent detection method |
| CN108875803A (en) * | 2018-05-30 | 2018-11-23 | 长安大学 | A kind of detection of harmful influence haulage vehicle and recognition methods based on video image |
| CN109934161A (en) * | 2019-03-12 | 2019-06-25 | 天津瑟威兰斯科技有限公司 | Vehicle identification and detection method and system based on convolutional neural network |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105373782A (en) * | 2015-11-16 | 2016-03-02 | 深圳市哈工大交通电子技术有限公司 | Method of automatically recognizing hazardous chemical vehicle from image or video |
| CN106709486A (en) * | 2016-11-11 | 2017-05-24 | 南京理工大学 | Automatic license plate identification method based on deep convolutional neural network |
| CN111353555A (en) * | 2020-05-25 | 2020-06-30 | 腾讯科技（深圳）有限公司 | Label detection method and device and computer readable storage medium |
| CN111523527B (en) * | 2020-07-02 | 2020-10-27 | 平安国际智慧城市科技股份有限公司 | Special transport vehicle monitoring method and device, medium and electronic equipment |
| CN111859872A (en) * | 2020-07-07 | 2020-10-30 | 中国建设银行股份有限公司 | Text labeling method and device |
| CN111898502A (en) * | 2020-07-20 | 2020-11-06 | 北京格灵深瞳信息技术有限公司 | Dangerous goods vehicle identification method and device, computer storage medium and electronic equipment |
| CN111860690A (en) * | 2020-07-31 | 2020-10-30 | 河北交投智能交通技术有限责任公司 | Dangerous goods vehicle detection and identification method based on deep learning |
| CN112200231B (en) * | 2020-09-29 | 2024-04-30 | 深圳市信义科技有限公司 | Dangerous goods vehicle identification method, system, device and medium |
-
2021
- 2021-09-06 CN CN202111035738.8A patent/CN113469158B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107886731A (en) * | 2017-11-03 | 2018-04-06 | 武汉元鼎创天信息科技有限公司 | A kind of illegal operation Vehicular intelligent detection method |
| CN108875803A (en) * | 2018-05-30 | 2018-11-23 | 长安大学 | A kind of detection of harmful influence haulage vehicle and recognition methods based on video image |
| CN109934161A (en) * | 2019-03-12 | 2019-06-25 | 天津瑟威兰斯科技有限公司 | Vehicle identification and detection method and system based on convolutional neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113469158A (en) | 2021-10-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113469158B (en) | Method and system for identifying illegal hazardous chemical substance transport vehicle based on convolutional neural network | |
| WO2020124247A1 (en) | Automated inspection system and associated method for assessing the condition of shipping containers | |
| CN106707293A (en) | Obstacle recognition method and device for vehicles | |
| US20190139262A1 (en) | Method and device for vehicle identification | |
| CN111797829A (en) | License plate detection method and device, electronic equipment and storage medium | |
| CN111460927B (en) | Method for extracting structured information of house property evidence image | |
| Ahmadi et al. | An integrated machine learning model for automatic road crack detection and classification in urban areas | |
| CN111626295B (en) | Training method and device for license plate detection model | |
| CN111325769A (en) | Target object detection method and device | |
| CN110309843B (en) | Automatic identification method for multiple types of components in power equipment image | |
| CN113052295B (en) | Training method of neural network, object detection method, device and equipment | |
| CN107194393A (en) | A kind of method and device for detecting Provisional Number Plate | |
| CN112749653A (en) | Pedestrian detection method, device, electronic equipment and storage medium | |
| CN111881958A (en) | License plate classification recognition method, device, equipment and storage medium | |
| Han et al. | Vision-based crack detection of asphalt pavement using deep convolutional neural network | |
| CN115995056A (en) | Automatic bridge disease identification method based on deep learning | |
| CN107316296A (en) | A kind of method for detecting change of remote sensing image and device based on logarithmic transformation | |
| CN111402185A (en) | Image detection method and device | |
| CN111160142A (en) | Certificate bill positioning detection method based on numerical prediction regression model | |
| CN116129100A (en) | Truck part positioning and fault identifying method, device, equipment and medium | |
| CN115512098A (en) | Electronic bridge inspection system and inspection method | |
| CN114219073A (en) | Method and device for determining attribute information, storage medium and electronic device | |
| CN113569829A (en) | Container coding data identification method and system | |
| Shin et al. | Visualization for explanation of deep learning-based defect detection model using class activation map | |
| CN117272124B (en) | Chemical classification method, system, equipment and storage medium based on energy level |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |